Stream: netherlands
Topic: Text/String fields
Yuri Bonanno (Jan 10 2019 at 12:42):
I am working on a model that has the structure of MedicalDevice HCIM to contain the data in parsed XML files, and I found something I did not quite know how to handle. As can be seen in the HCIM, both ProductDescription and Comment have kardinality (0..1). While on MedicalDevice Simplifier it shows that ,for both, an indefinite amount can be defined.
My solution right now is to concatenate all the strings that can possibly found and puts them into the (0..1) field. This way, if multiple Strings are changed into a single String. So this way it fits into the by Nictiz defined parameters.
Is this a good way to go about it? Should I instead return an error when multiple comments are found? Or just parse the first one and then the others?
Alexander Henket (Jan 11 2019 at 13:04):
You are referring to the 2018 prerelease version of the HCIM. We only support MedicalDevice 2017. Same issue applies there too though.
The HCIM is a compatibility model and does not suggest to be a maximum number of properties for a given concept. The maximum number of properties for any concept is potentially infinite. E.g. any specialist will likely have a hundreds more properties that are not covered by the HCIM.
That being said: if a system supports 2 descriptions or 2 comments for a medical device and exports those into FHIR, a receiving system is expected to be compatible with 1 of each based on the HCIM. If however the receiving system also supports 2 discrete descriptions/comments, it might import them as received.
If the system only supports what the HCIM suggested (1 each), it could concatenate the descriptions and the comments into one each, or be prepared to loose every second description/comment and beyond. Sending an error is not an option because the sending system is not doing anything wrong.
In practice I hope we will find that initially most systems will work like yours: implement the HCIM as well as possible. The profiles try to cater for the real world that is more detailed/complex than any given HCIM suggests. When systems are at least interoperable at HCIM level, we have done a lot already.
Last updated: Apr 12 2022 at 19:14 UTC
